Support trailing commas in expression lists

ClickHouse allows trailing commas before clauses like FROM, WHERE, etc.
For example: SELECT a, b, FROM table

Add isClauseKeyword function to detect when a token is a clause keyword
that should terminate an expression list, vs when it's being used as
an identifier (which ClickHouse allows for many keywords).

The detection is context-aware - keywords followed by (, [, or = are
treated as expression continuations rather than clause terminators.

Author: claude

parser/expression.go

@@ -99,6 +99,10 @@ func (p *Parser) parseExpressionList() []ast.Expression {
 
 	for p.currentIs(token.COMMA) {
 		p.nextToken()
+		// Handle trailing commas by checking if next token is a clause keyword
+		if p.isClauseKeyword() {
+			break
+		}
 		expr := p.parseExpression(LOWEST)
 		if expr != nil {
 			// Handle implicit alias (identifier without AS)
@@ -110,6 +114,29 @@ func (p *Parser) parseExpressionList() []ast.Expression {
 	return exprs
 }
 
+// isClauseKeyword returns true if the current token is a SQL clause keyword
+// that should terminate an expression list (used for trailing comma support)
+func (p *Parser) isClauseKeyword() bool {
+	switch p.current.Token {
+	// Only these tokens are unambiguously clause terminators
+	case token.RPAREN, token.SEMICOLON, token.EOF:
+		return true
+	// FROM is a clause keyword unless followed by ( or [ (function/index access)
+	case token.FROM:
+		return !p.peekIs(token.LPAREN) && !p.peekIs(token.LBRACKET)
+	// These keywords can be used as identifiers in ClickHouse
+	// Only treat as clause keywords if NOT followed by expression-like tokens
+	case token.WHERE, token.GROUP, token.HAVING, token.ORDER, token.LIMIT:
+		// If followed by comma, it's likely an identifier in a list
+		return !p.peekIs(token.LPAREN) && !p.peekIs(token.LBRACKET) && !p.peekIs(token.COMMA) && !p.peekIs(token.RPAREN)
+	case token.INTO, token.SETTINGS, token.FORMAT:
+		return !p.peekIs(token.LPAREN) && !p.peekIs(token.LBRACKET) && !p.peekIs(token.EQ) && !p.peekIs(token.COMMA) && !p.peekIs(token.RPAREN)
+	// UNION, EXCEPT, INTERSECT, OFFSET are often used as identifiers
+	// Don't treat them as clause terminators to avoid breaking valid code
+	}
+	return false
+}
+
 // parseGroupingSets parses GROUPING SETS ((a), (b), (a, b))
 func (p *Parser) parseGroupingSets() []ast.Expression {
 	var exprs []ast.Expression

parser/testdata/02354_vector_search_binary_quantization/metadata.json

@@ -1,5 +1 @@
-{
-  "explain_todo": {
-    "stmt22": true
-  }
-}
+{}

parser/testdata/02813_optimize_lazy_materialization/metadata.json

@@ -1,6 +1 @@
-{
-  "explain_todo": {
-    "stmt136": true,
-    "stmt141": true
-  }
-}
+{}

parser/testdata/02842_mutations_replace_non_deterministic/metadata.json

@@ -1,6 +1 @@
-{
-  "explain_todo": {
-    "stmt28": true,
-    "stmt42": true
-  }
-}
+{}

parser/testdata/03008_deduplication_remote_insert_select/metadata.json

@@ -1,5 +1 @@
-{
-  "explain_todo": {
-    "stmt9": true
-  }
-}
+{}

parser/testdata/03035_argMinMax_numeric_non_extreme_bug/metadata.json

@@ -1,5 +1 @@
-{
-  "explain_todo": {
-    "stmt3": true
-  }
-}
+{}

parser/testdata/03040_dynamic_type_alters_2_compact_merge_tree/metadata.json

@@ -1,6 +1,5 @@
 {
   "explain_todo": {
-    "stmt24": true,
-    "stmt30": true
+    "stmt24": true
   }
 }

parser/testdata/03040_dynamic_type_alters_2_wide_merge_tree/metadata.json

@@ -1,6 +1,5 @@
 {
   "explain_todo": {
-    "stmt25": true,
-    "stmt32": true
+    "stmt25": true
   }
 }

parser/testdata/03047_on_fly_mutations_non_deterministic_replace/metadata.json

@@ -1,6 +1 @@
-{
-  "explain_todo": {
-    "stmt36": true,
-    "stmt58": true
-  }
-}
+{}

parser/testdata/03156_tuple_map_low_cardinality/metadata.json

@@ -1,10 +1 @@
-{
-  "explain_todo": {
-    "stmt11": true,
-    "stmt12": true,
-    "stmt13": true,
-    "stmt4": true,
-    "stmt5": true,
-    "stmt6": true
-  }
-}
+{}